Skip to main content
Soft packages, cartons, and sacks can deform, adhere to one another, or sit at an angle, making recognition, grasping, and sequence planning difficult for a robot arm. This solution targets automated depalletizing of plastic pellets, fertilizer, food products, and similar materials.

​
Solution overview

An RGB-D camera captures color and point-cloud data from the stack. A camera-side algorithm identifies target packages and their positions. The result can generate a grasp sequence and provide the calibration data needed for robot-arm coordinate conversion.

​
Solution characteristics

3D recognition

Combine surface texture and point-cloud data to identify the shape and position of irregular bags and cartons.

On-camera processing

Run the depalletizing algorithm inside the camera to reduce dependence on an external industrial computer.

Stack-pattern adaptation

Adapt to stack patterns such as six-flower, five-flower, and ring arrangements.

Grasp-sequence planning

Generate a grasp sequence from package positions to guide the robot arm package by package.
3D soft-package depalletizing recognition result

​
Calibration and integration

The system provides a four-point calibration method to calculate the transformation matrix between the camera and robot arm. It has been adapted to industrial robot brands including ABB, KUKA, FANUC, AUBO, and DOBOT and supports TCP/IP communication. Confirm the actual integration scope through project validation.
Four-point calibration for soft-package depalletizing
Soft-package depalletizing system integrated with a robot